基于Qt的yolov5工程

Yolov5Qt工程
main.cpp

#include "mainwindow.h"

#include <QApplication>

int main(int argc, char *argv[])
{
    
    QApplication a(argc, argv);
    MainWindow w;
    w.show();
    return a.exec();
}

mainwindow.cpp

#include "mainwindow.h"
#include "ui_mainwindow.h"



MainWindow::MainWindow(QWidget *parent)
    : QMainWindow(parent)
    , ui(new Ui::MainWindow)
{
    
    ui->setupUi(this);
    setWindowTitle(QStringLiteral("YoloV5目标检测软件"));

    timer = new QTimer(this);
    timer->setInterval(33);
    connect(timer,SIGNAL(timeout()),this,SLOT(readFrame()));
    ui->startdetect->setEnabled(false);
    ui->stopdetect->setEnabled(false);
    Init();
}

MainWindow::~MainWindow()
{
    

    capture->release();
    delete capture;
    delete [] yolo_nets;
    delete yolov5;
    delete ui;
}

void MainWindow::Init()
{
    
    capture = new cv::VideoCapture();
    yolo_nets = new NetConfig[4]{
    
                                {
    0.5, 0.5, 0.5, "yolov5s"},
                                {
    0.6, 0.6, 0.6, "yolov5m"},
                                {
    0.65, 0.65, 0.65, "yolov5l"},
                                {
    0.75, 0.75, 0.75, "yolov5x"}
                            };
    conf = yolo_nets[0];
    yolov5 = new YOLOV5();
    yolov5->Initialization(conf);
    ui->textEditlog->append(QStringLiteral("默认模型类别：yolov5s args: %1 %2 %3")
                            .arg(conf.nmsThreshold)
                            .arg(conf.objThreshold)
                            .arg(conf.confThreshold));
}

void MainWindow::readFrame()
{
    
    cv::Mat frame;
    capture->read(frame);
    if (frame.empty()) return;

    auto start = std::chrono::steady_clock::now();
    yolov5->detect(frame);
    auto end = std::chrono::steady_clock::now();
    std::chrono::duration<double, std::milli> elapsed = end - start;
    ui->textEditlog->append(QString("cost_time: %1 ms").arg(elapsed.count()));
    cv::cvtColor(frame, frame, cv::COLOR_BGR2RGB);
    QImage rawImage = QImage((uchar*)(frame.data),frame.cols,frame.rows,frame.step,QImage::Format_RGB888);
    ui->label->setPixmap(QPixmap::fromImage(rawImage));
}

void MainWindow::on_openfile_clicked()
{
    
    QString filename = QFileDialog::getOpenFileName(this,QStringLiteral("打开文件"),".","*.mp4 *.avi;;*.png *.jpg *.jpeg *.bmp");
    if(!QFile::exists(filename)){
    
        return;
    }
    ui->statusbar->showMessage(filename);

    QMimeDatabase db;
    QMimeType mime = db.mimeTypeForFile(filename);
    if (mime.name().startsWith("image/")) {
    
        cv::Mat src = cv::imread(filename.toLatin1().data());
        if(src.empty()){
    
            ui->statusbar->showMessage("图像不存在！");
            return;
        }
        cv::Mat temp;
        if(src.channels()==4)
            cv::cvtColor(src,temp,cv::COLOR_BGRA2RGB);
        else if (src.channels()==3)
            cv::cvtColor(src,temp,cv::COLOR_BGR2RGB);
        else
            cv::cvtColor(src,temp,cv::COLOR_GRAY2RGB);

        auto start = std::chrono::steady_clock::now();
        yolov5->detect(temp);
        auto end = std::chrono::steady_clock::now();
        std::chrono::duration<double, std::milli> elapsed = end - start;
        ui->textEditlog->append(QString("cost_time: %1 ms").arg(elapsed.count()));
        QImage img = QImage((uchar*)(temp.data),temp.cols,temp.rows,temp.step,QImage::Format_RGB888);
        ui->label->setPixmap(QPixmap::fromImage(img));
        ui->label->resize(ui->label->pixmap()->size());
        filename.clear();
    }else if (mime.name().startsWith("video/")) {
    
        capture->open(filename.toLatin1().data());
        if (!capture->isOpened()){
    
            ui->textEditlog->append("fail to open MP4!");
            return;
        }
        IsDetect_ok +=1;
        if (IsDetect_ok ==2)
            ui->startdetect->setEnabled(true);
        ui->textEditlog->append(QString::fromUtf8("Open video: %1 succesfully!").arg(filename));

        //获取整个帧数QStringLiteral
        long totalFrame = capture->get(cv::CAP_PROP_FRAME_COUNT);
        int width = capture->get(cv::CAP_PROP_FRAME_WIDTH);
        int height = capture->get(cv::CAP_PROP_FRAME_HEIGHT);
        ui->textEditlog->append(QStringLiteral("整个视频共 %1 帧, 宽=%2 高=%3 ").arg(totalFrame).arg(width).arg(height));
        ui->label->resize(QSize(width, height));

        //设置开始帧()
        long frameToStart = 0;
        capture->set(cv::CAP_PROP_POS_FRAMES, frameToStart);
        ui->textEditlog->append(QStringLiteral("从第 %1 帧开始读").arg(frameToStart));

        //获取帧率
        double rate = capture->get(cv::CAP_PROP_FPS);
        ui->textEditlog->append(QStringLiteral("帧率为: %1 ").arg(rate));
    }
}

void MainWindow::on_loadfile_clicked()
{
    
    QString onnxFile = QFileDialog::getOpenFileName(this,QStringLiteral("选择模型"),".","*.onnx");
    if(!QFile::exists(onnxFile)){
    
        return;
    }
    ui->statusbar->showMessage(onnxFile);
    if (!yolov5->loadModel(onnxFile.toLatin1().data())){
    
        ui->textEditlog->append(QStringLiteral("加载模型失败！"));
        return;
    }
    IsDetect_ok +=1;
    ui->textEditlog->append(QString::fromUtf8("Open onnxFile: %1 succesfully!").arg(onnxFile));
    if (IsDetect_ok ==2)
        ui->startdetect->setEnabled(true);
}

void MainWindow::on_startdetect_clicked()
{
    
    timer->start();
    ui->startdetect->setEnabled(false);
    ui->stopdetect->setEnabled(true);
    ui->openfile->setEnabled(false);
    ui->loadfile->setEnabled(false);
    ui->comboBox->setEnabled(false);
    ui->textEditlog->append(QStringLiteral("=======================\n"
                                           " 开始检测\n"
                                           "=======================\n"));
}

void MainWindow::on_stopdetect_clicked()
{
    
    ui->startdetect->setEnabled(true);
    ui->stopdetect->setEnabled(false);
    ui->openfile->setEnabled(true);
    ui->loadfile->setEnabled(true);
    ui->comboBox->setEnabled(true);
    timer->stop();
    ui->textEditlog->append(QStringLiteral("======================\n"
                                           " 停止检测\n"
                                           "======================\n"));
}

void MainWindow::on_comboBox_activated(const QString &arg1)
{
    
    if (arg1.contains("s")){
    
        conf = yolo_nets[0];
    }else if (arg1.contains("m")) {
    
        conf = yolo_nets[1];
    }else if (arg1.contains("l")) {
    
        conf = yolo_nets[2];
    }else if (arg1.contains("x")) {
    
        conf = yolo_nets[3];}
    yolov5->Initialization(conf);
    ui->textEditlog->append(QStringLiteral("使用模型类别：%1 args: %2 %3 %4")
                            .arg(arg1)
                            .arg(conf.nmsThreshold)
                            .arg(conf.objThreshold)
                            .arg(conf.confThreshold));
}

yolov5.h

#ifndef YOLOV5_H
#define YOLOV5_H
#include <opencv2/opencv.hpp>
#include <opencv2/dnn.hpp>
#include <opencv2/core/cuda.hpp>
#include <fstream>
#include <sstream>
#include <iostream>
#include <exception>
#include <QMessageBox>


struct NetConfig
{
    
    float confThreshold; // class Confidence threshold
    float nmsThreshold;  // Non-maximum suppression threshold
    float objThreshold;  //Object Confidence threshold
    std::string netname;
};

class YOLOV5
{
    
public:
    YOLOV5(){
    }  //构造函数
    void Initialization(NetConfig conf);
    bool loadModel(const char* onnxfile);
    void detect(cv::Mat& frame);
private:
    const float anchors[3][6] = {
    {
    10.0, 13.0, 16.0, 30.0, 33.0, 23.0}, {
    30.0, 61.0, 62.0, 45.0, 59.0, 119.0},{
    116.0, 90.0, 156.0, 198.0, 373.0, 326.0}};
    const float stride[3] = {
     8.0, 16.0, 32.0 };
    std::string classes[80] = {
    "person", "bicycle", "car", "motorbike", "aeroplane", "bus",
                              "train", "truck", "boat", "traffic light", "fire hydrant",
                              "stop sign", "parking meter", "bench", "bird", "cat", "dog",
                              "horse", "sheep", "cow", "elephant", "bear", "zebra", "giraffe",
                              "backpack", "umbrella", "handbag", "tie", "suitcase", "frisbee",
                              "skis", "snowboard", "sports ball", "kite", "baseball bat",
                              "baseball glove", "skateboard", "surfboard", "tennis racket",
                              "bottle", "wine glass", "cup", "fork", "knife", "spoon", "bowl",
                              "banana", "apple", "sandwich", "orange", "broccoli", "carrot",
                              "hot dog", "pizza", "donut", "cake", "chair", "sofa", "pottedplant",
                              "bed", "diningtable", "toilet", "tvmonitor", "laptop", "mouse",
                              "remote", "keyboard", "cell phone", "microwave", "oven", "toaster",
                              "sink", "refrigerator", "book", "clock", "vase", "scissors",
                              "teddy bear", "hair drier", "toothbrush"};
    const int inpWidth = 640;
    const int inpHeight = 640;
    float confThreshold;
    float nmsThreshold;
    float objThreshold;

    //========= test =========
    std::vector<int> blob_sizes{
     1, 3, 640, 640};
    cv::Mat blob = cv::Mat(blob_sizes, CV_32FC1, cv::Scalar(0.0));

    //========== pro ========
    //cv::Mat blob;
    std::vector<cv::Mat> outs;
    std::vector<int> classIds;
    std::vector<float> confidences;
    std::vector<cv::Rect> boxes;
    std::vector<int> indices;
    cv::dnn::Net net;
    void drawPred(int classId, float conf, int left, int top, int right, int bottom, cv::Mat& frame);
    void sigmoid(cv::Mat* out, int length);
};

static inline float sigmoid_x(float x)
{
    
    return static_cast<float>(1.f / (1.f + exp(-x)));
}
#endif // YOLOV5_H

yolov5.cpp

#include "yolov5.h"
using namespace std;
using namespace cv;



void YOLOV5::Initialization(NetConfig conf)
{
    
    this->confThreshold = conf.confThreshold;
    this->nmsThreshold = conf.nmsThreshold;
    this->objThreshold = conf.objThreshold;
    classIds.reserve(20);
    confidences.reserve(20);
    boxes.reserve(20);
    outs.reserve(3);
    indices.reserve(20);
}

bool YOLOV5::loadModel(const char *onnxfile)
{
    

// try {
    
// this->net = cv::dnn::readNetFromONNX(onnxfile);
// int device_no = cv::cuda::getCudaEnabledDeviceCount();
// if (device_no==1){
    
// this->net.setPreferableBackend(cv::dnn::DNN_BACKEND_CUDA);
// this->net.setPreferableTarget(cv::dnn::DNN_TARGET_CUDA);
// }else{
    
// QMessageBox::information(NULL,"warning",QStringLiteral("正在使用CPU推理！\n"),QMessageBox::Yes,QMessageBox::Yes);
// }
// return true;
// } catch (exception& e) {
    
// QMessageBox::critical(NULL,"Error",QStringLiteral("模型加载出错，请检查重试！\n %1").arg(e.what()),QMessageBox::Yes,QMessageBox::Yes);
// return false;
// }
    this->net = cv::dnn::readNetFromONNX(onnxfile);
    this->net.setPreferableBackend(cv::dnn::DNN_BACKEND_DEFAULT);
    this->net.setPreferableTarget(cv::dnn::DNN_TARGET_CPU);



// if(1 == cv::cuda::getCudaEnabledDeviceCount()){
    
// this->net.setPreferableBackend(cv::dnn::DNN_BACKEND_CUDA);
// this->net.setPreferableTarget(cv::dnn::DNN_TARGET_CUDA);
// }
// this->net.setPreferableBackend(cv::dnn::DNN_BACKEND_DEFAULT);
// this->net.setPreferableTarget(cv::dnn::DNN_TARGET_CPU);
// this->net.setPreferableBackend(cv::dnn::DNN_BACKEND_INFERENCE_ENGINE);
// this->net.setPreferableTarget(cv::dnn::DNN_TARGET_CPU);
}

void YOLOV5::detect(cv::Mat &frame)
{
    
    cv::dnn::blobFromImage(frame, blob, 1 / 255.0, Size(this->inpWidth, this->inpHeight), Scalar(0, 0, 0), true, false);
    this->net.setInput(blob);
    this->net.forward(outs, this->net.getUnconnectedOutLayersNames());

    /generate proposals
    classIds.clear();
    confidences.clear();
    boxes.clear();
    float ratioh = (float)frame.rows / this->inpHeight, ratiow = (float)frame.cols / this->inpWidth;
    int n = 0, q = 0, i = 0, j = 0, nout = 8 + 5, c = 0;
    for (n = 0; n < 3; n++)   ///尺度
    {
    
        int num_grid_x = (int)(this->inpWidth / this->stride[n]);
        int num_grid_y = (int)(this->inpHeight / this->stride[n]);
        int area = num_grid_x * num_grid_y;
        this->sigmoid(&outs[n], 3 * nout * area);
        for (q = 0; q < 3; q++)    ///anchor数
        {
    
            const float anchor_w = this->anchors[n][q * 2];
            const float anchor_h = this->anchors[n][q * 2 + 1];
            float* pdata = (float*)outs[n].data + q * nout * area;
            for (i = 0; i < num_grid_y; i++)
            {
    
                for (j = 0; j < num_grid_x; j++)
                {
    
                    float box_score = pdata[4 * area + i * num_grid_x + j];
                    if (box_score > this->objThreshold)
                    {
    
                        float max_class_socre = 0, class_socre = 0;
                        int max_class_id = 0;
                        for (c = 0; c < 80; c++)  get max socre
                        {
    
                            class_socre = pdata[(c + 5) * area + i * num_grid_x + j];
                            if (class_socre > max_class_socre)
                            {
    
                                max_class_socre = class_socre;
                                max_class_id = c;
                            }
                        }

                        if (max_class_socre > this->confThreshold)
                        {
    
                            float cx = (pdata[i * num_grid_x + j] * 2.f - 0.5f + j) * this->stride[n];  ///cx
                            float cy = (pdata[area + i * num_grid_x + j] * 2.f - 0.5f + i) * this->stride[n];   ///cy
                            float w = powf(pdata[2 * area + i * num_grid_x + j] * 2.f, 2.f) * anchor_w;   ///w
                            float h = powf(pdata[3 * area + i * num_grid_x + j] * 2.f, 2.f) * anchor_h;  ///h

                            int left = (cx - 0.5*w)*ratiow;
                            int top = (cy - 0.5*h)*ratioh;   ///坐标还原到原图上

                            classIds.push_back(max_class_id);
                            confidences.push_back(max_class_socre);
                            boxes.push_back(Rect(left, top, (int)(w*ratiow), (int)(h*ratioh)));
                        }
                    }
                }
            }
        }
    }

    // Perform non maximum suppression to eliminate redundant overlapping boxes with
    // lower confidences
    indices.clear();
    cv::dnn::NMSBoxes(boxes, confidences, this->confThreshold, this->nmsThreshold, indices);
    for (size_t i = 0; i < indices.size(); ++i)
    {
    
        int idx = indices[i];
        Rect box = boxes[idx];
        this->drawPred(classIds[idx], confidences[idx], box.x, box.y,
            box.x + box.width, box.y + box.height, frame);
    }
}

void YOLOV5::drawPred(int classId, float conf, int left, int top, int right, int bottom, Mat &frame)
{
    
    rectangle(frame, Point(left, top), Point(right, bottom), Scalar(0, 0, 255), 3);
    string label = format("%.2f", conf);
    label = this->classes[classId] + ":" + label;

    int baseLine;
    Size labelSize = getTextSize(label, FONT_HERSHEY_SIMPLEX, 0.5, 1, &baseLine);
    top = max(top, labelSize.height);
    putText(frame, label, Point(left, top), FONT_HERSHEY_SIMPLEX, 0.75, Scalar(0, 255, 0), 1);
}

void YOLOV5::sigmoid(Mat *out, int length)
{
    
    float* pdata = (float*)(out->data);
    int i = 0;
    for (i = 0; i < length; i++)
    {
    
        pdata[i] = 1.0 / (1 + expf(-pdata[i]));
    }
}